On 5/5/16 2:08 PM, Piotr Grzybowski wrote:
> Hey,
>
> confirmed, this is what happens by construction.
> this one is going to be tricky, expansion happens on the shell line, way
> before it comes to execution. I do not see reparsing again at execute_command
> even if we had means, which we dont, to store the information from history
> expansion to know what to execute and what not. Re-tokenizing by rules of
> history expansion anywhere outside histexpand seems out of place as much as
> implementing shell lexer (again) in histexpand. Maybe splitting in
> history_expand and returning tokens with options, that are later put together
> for execution? Either way seems a bit large.
> Currently the design indicates: :p applies to the whole line.
Yes, the behavior as inherited from csh and tcsh should be that :p in any
history expansion on the line will apply to the entire line. It's an
easy one-character fix.
--
``The lyf so short, the craft so long to lerne.'' - Chaucer
``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, ITS, CWRU c...@case.edu http://cnswww.cns.cwru.edu/~chet/
